Working Paper: NBER ID: w3368
Authors: Karl E. Case; Robert J. Shiller
Abstract: The U. S. market for homes appears not to be efficient. A number of information variables predict housing price changes and excess returns of housing relative to debt over the succeeding year. Price changes observed over one year tend to continue for one more year in the same direction. Construction cost divided by price, the change in per capita real income, the change in adult population are all positively related to price changes or excess returns over the subsequent year. \nThe results are based on time-series cross section regressions with quarterly data 1970-1 to 1987-3 and for cities Atlanta, Chicago, Dallas, and San Francisco.
Keywords: housing market; price forecasting; excess returns
JEL Codes: R31; G12
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| construction costs (L74) | housing prices (R31) |
| changes in per capita real income (O49) | housing prices (R31) |
| changes in adult population (J11) | housing prices (R31) |
| construction costs (L74) | excess returns (D46) |
| changes in per capita real income (O49) | excess returns (D46) |
| changes in adult population (J11) | excess returns (D46) |
| price changes (P22) | future price changes (G13) |